The first time I saw these photovoltaic opto's used was for a high side drive of a MOS-FET gate. This is a quick and dirty way of using an N-channel part without needing an extra power supply. As you observe, the gate signal can't change very fast so PWM is not very practical.Yes, it would seem that you could get a few volts of output voltage if you have high output load resistance ( a few hundred kΩ or more).
Obviously the frequency response would be limited at those high impedances but it should work for low frequencies, depending upon the stray output capacitance, or DC.
The more I look at that top left graph the less I understand. At least the graph below it makes sense to me.The graphs on page 5 only reflect the coupling with an applied voltage at the output.